I would consider -permute- for permutation tests based on Monte Carlo
simulations. What assumptions can you make, e.g., do you have
independent samples (i.e. unmatched data)?
Anders Alexandersson
andersalex@gmail.com
Privately, Hema replied:
The "non-randomised data" refers to both the sample and the treatment
assignment.
On 4/20/06, Anders Alexandersson <andersalex@gmail.com> wrote:
Does "non-randomised data" here refer to the sample and/or to the
treatment assignment?
On 4/20/06, Hema Mistry <Hema.Mistry@brunel.ac.uk> wrote:
> I was wondering whether you can provide me with some advice or point me
> in the right
> direction. I am trying to find methods which can deal with data that
> is non-randomised
> and suffers from selection bias. After searching various databases etc
> I have come up
> with the following methods:
> 1) Regression analyses
> 2) Propensity score - matching, stratification, regression,
> classification trees
> 3) Instrumental variables
> 4) Sample selection models
> 5) Two-part models
> 6) Inverse probability weighting
>
> Before I start using these methods in various datasets I was just
> wondering whether
> users are aware of any other methods which I have not identified?
> Can you recommend any good text books or key people that maybe I should
> contact?
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/